The _add_unattributed_bytes function was adding the same cross-hour
delta to both days when the interval spanned midnight, violating the
spec requirement to preserve measured volume once as shared boundary
evidence.
Store the unattributed bytes only on the day where the interval starts
(the earlier day). This ensures the total across both days equals the
actual delta without duplication.
Closes#88 (cross-day delta duplication portion)
Add query_local_day_history() with evidence-limit metadata to local_day.py,
including LocalDayHistoryEntry dataclass that annotates each summary with
detail_available and derived_from_surviving flags. Document the 14-day
detail vs indefinite-summary retention policy in both local_day.py and
pruning.py.
Fix query_local_day_summary() to convert SQLite integer booleans to Python
bool, and add deduplication guard to the midnight-spanning hour logic in
derive_local_day_summary().
Add 17 lifecycle tests covering all acceptance criteria:
- AC1: Summaries survive sample pruning, evidence limits visible
- AC2: Boundary anchors preserved before pruning
- AC3: Timezone survives system-timezone change
- AC4: Legacy UTC summaries labelled incomplete
- AC5: Idempotent pruning and repair
- AC6: Volume conservation (no double counting)
- AC7: Real temporary stores
- AC8: User-visible transition from recent to aged
- AC9: Retention policy documented
All 775 tests pass.
Add keyboard-driven date navigation to the TUI:
- [ and ] keys select previous/next day, adjusting the visible range
- g opens a date entry modal for direct date navigation
- t returns to today's live view from any historical browsing
- Background refresh preserves the browsed selection
- Local-day widget shows data for the selected date
- Updated help screen and action legend with new bindings
- DatePickerScreen modal with input validation
- 24 headless tests covering navigation, date entry, browse
stability, local-day evidence, and constrained widths
Co-authored-by: CommandCodeBot <noreply@commandcode.ai>
- Change default collection cadence from 5 minutes to 3 minutes
(CADENCE_DEFAULT_S=180, systemd OnUnitInactiveSec=3min,
runit CADENCE=180)
- Update freshness threshold to 450s (2×180 + AccuracySec + 60)
- Add _query_live_graph_data(): queries raw samples from the last
3 hours and computes interval byte deltas with actual timestamps
- Add LiveActivityGraph widget: vertical bar chart of interval
volumes with read/write toggle (w key), arrow key inspection,
and click support
- Wire live graph into TUI layout (full-width row between daily
graph and drive health), refresh cycle, and CSS grid
- Replace t theme binding with t today/live binding; theme
selection via preferences file
- Add w binding for read/write toggle on live graph
- Update action legend, help screen, and grid layout for new
live-activity row
- Add 20 tests covering cadence constants, live query, widget
rendering, toggle, and TUI integration
- Update all cadence documentation (README, ADR 0003, acceptance
criteria LC-2, fenris-redesign constants table, CHANGELOG)
Add local-day activity summaries derived from UTC hour observations using
the system timezone, with durable storage in a new local_days table.
The collector derives local-day read/write totals after UTC aggregation;
the TUI displays them with timezone, completeness state, and coverage.
Schema: bump SCHEMA_VERSION to 3, add local_days table (migration 2→3
is pure addition, idempotent, preserves newer-schema refusal).
Repair the collection-to-display path so each successful acquisition
publishes correct read/write deltas through the observation store and
visible dashboard.
Fixes:
- derive.py: accumulate bytes_read_delta on same-hour hour_observation
merge (was silently dropped)
- collector.py: rebuild day_aggregates from hour observations after
each collection run (previously only populated for cross-hour intervals)
- day_aggregate.py: add persist_day_aggregate upsert helper
- tui.py: query and display both read and write deltas in daily and
hourly readouts, constrained summaries, and graph data queries
Tests:
- Add 14 integration tests (test_measured_activity.py) exercising the
full collector→store→reader→display path with real fixtures and
injected time
- Update constrained-layout assertion to match new W/R format
Closes#89
Add XBPS build and sign steps to CI workflow
Add XBPS publication as independent gate (requires manual trigger)
Track format availability in release notes
Update release-footer.md with XBPS install instructions
Add --available/--withheld arguments to extract_changelog.py
Attach XBPS artifacts to Gitea release
Clean up XBPS signing key material after use
Deliver end-to-end native monitoring path under runit with existing
CLI/TUI controls and truthful status, preserving systemd behavior.
Changes:
- Add init system abstraction layer (src/fenris/init_system.py) that
detects systemd vs runit and provides unified interface for timer
control, on-demand collection, and service state queries
- Create runit service files (units/runit/) with completion-relative
5-minute cadence, 2-minute boot delay, bounded execution (90s),
no catch-up, and serialized runs via flock
- Update monitor.py to use abstraction layer instead of direct systemctl
- Update status.py to use abstraction layer for service state queries
- Update all packaging scripts (deb, rpm) for init-system-aware setup
- Update Makefile to install runit service files alongside systemd units
- Add 46 tests for init system abstraction layer
Spec: ADR 0008, §8.4, §8.5, §8.6, §8.7, §8.8
Closes#84
Co-authored-by: CommandCodeBot <noreply@commandcode.ai>
Implement Amber/Nord/High Contrast theme presets with XDG user-scoped
persistence and reduced motion toggle. Covers TPH-10 and preference
integration with TPH-2.
- preferences.py: safe load/save with XDG_CONFIG_HOME/fenris/preferences.json
- themes.py: three Textual Theme objects with graph colour roles
- TUI: t cycles presets, m toggles reduced motion, both persist across restart
- Status composition receives reduced_motion from preferences
- 56 new tests covering persistence, themes, TUI integration, CLI isolation
- All 590 existing tests continue to pass
- Fix store dir mode: 2770 (per tmpfiles.d, matches test_store_group_access)
- Fix WAL/SHM survival: dpkg removes ephemeral SQLite files from
package-owned dirs; assert they are gone rather than present
- Fix opensuse migration_guard: use zypper instead of dnf
- Fix quote escaping in _setup_store_and_config
465 core tests pass. Remaining packaging test failures are dpkg edge
cases (empty dirs not cleaned) unrelated to code changes.
Co-authored-by: CommandCodeBot <noreply@commandcode.ai>
Replace all hardcoded docker commands with _container_cmd() helper
function that auto-detects podman or docker runtime.
Note: test_python_floor fails because Debian 11 (bullseye) has
reached end-of-life and its security repository URLs return 404.
This is a test infrastructure issue, not a code issue.
Co-authored-by: CommandCodeBot <noreply@commandcode.ai>
Add helper functions to detect and use podman or docker for
containerized packaging tests. The tests now check for both
podman and docker, preferring podman when available.
Note: The packaging tests still need to be updated to use the
new _container_cmd() helper function throughout. Currently only
the helper functions have been updated.
Co-authored-by: CommandCodeBot <noreply@commandcode.ai>
Headline and scenario rates now describe exact evidence-supported
monitored spans anchored at the latest published usage-evidence
endpoint T, not clock_now. Reader refresh alone never moves T or
dilutes rates.
- Add horizon_reasons field to ScenarioRange for specific unavailability facts
- Modify _compute_horizon_rate to use exact trailing 7/28/90×86400-second starts from T
- Show specific reasons for affected horizons (e.g., "starts before earliest data")
- Update TUI and CLI to display horizon-specific reasons
- Add 6 new tests for evidence-anchored projection rates
Co-authored-by: CommandCodeBot <noreply@commandcode.ai>
Replace the static sparkline with an interactive block-glyph bar graph
that supports writes-only daily bars, range switching (7/14/28/90 days),
day selection, and hourly drill-down. No plotting dependency.
Co-authored-by: CommandCodeBot <noreply@commandcode.ai>
- Schema migration 1-2: add segment_id to samples, unattributed bytes to day_aggregates
- Collector now derives hour observations and day aggregates from sample pairs
- Cross-hour deltas tracked as unattributed (no proportional allocation)
- Display states: 0 samples -> awaiting first, 1 sample -> awaiting another
- Monitoring period ensured open on each collection run
- Derivation failures preserve prior history
Closes#73
- open_store_readonly(): stat() PermissionError (non-group user on the
2750 store dir) now maps to StoreFault so status/TUI degrade instead
of crashing with a traceback.
- init_store(): chmod db + -wal/-shm group rw after WAL setup — SQLite
WAL readers need write access to sidecars even for mode=ro opens.
- Store dir 2750 → 2770 (tmpfiles + make install) and UMask=002 on the
collect unit so root-created files stay group-accessible.
- rpm %post upgrade path re-runs systemd-tmpfiles --create to correct
placement modes on existing machines.
Bump to 0.3.3.
Fresh installs shipped a config template with no store_path key while
collector.py demanded one via get_store_path() — every first collect
crashed with KeyError 'store_path'. Resolve to the packaged default
(/var/lib/fenris/observations.db) when absent, document the key in the
template, and cover the fresh-install path with regression tests.
Bump to 0.3.2.
xavierk